Ultra Low Latency - The Real Definition!


September 21, 2022

Ultra Low Latency - The Real Definition!

When it comes to live streaming, one factor that can affect the viewing experience of a live news interview is the cringing delay between the presenter and the reporter in the field. Is it possible to stream without delay?

The answer is simply no. There will always be delay in every part of the live streaming workflow. There is latency in the camera. There is latency with encoding when the video is compressed. There is the network latency between locations. There is latency when the video is decoded, and finally latency even within the monitor where a video is finally displayed. Even though the latency maybe small, there will always be some induced delay.

What is the lowest latency for video streaming?

It depends on many factors and the type of video streaming – whether it is camera to studio, studio to studio, studio to playout, or playout to a smart TV or workstation via public internet. Typically live streaming over cellular can take a few seconds, satellite streaming can take seconds, studio-to-studio with fiber can take milliseconds, studio to playout over a public network can take seconds. YouTube can be 30 seconds plus. With unstable public networks, latency maybe increased on purpose for buffering to improve quality and reliability.

main-visual-ultra-low-latency-h265

Do you have real examples of ultra low latency for live streaming?

The lowest latency for video contribution from a live camera to the studio, as in the case of a live news report, can be as low as 65ms over 4G using Soliton's Zao-X encoders. Typically the closest competitive offerings have a latency of 1-2 seconds. The closest, claim a latency of 200ms over a private cellular networks where there is zero contention. Soliton over public 5G and wired connections can be as low as 35ms, even over public internet connections. Soliton has also tested over Starlink.

 What does ultra low latency mean?

There is no universally recognized definition of ultra low latency. For example Wowza claim it as under  a second, Cisco claim it in nanoseconds, but admittedly the latter is not for video streaming. With video streaming, given that ultra low latency is seen as the lowest latency that can be physically achieved, anything within a few frames of video is defined currently as ultra-low latency, meaning under 100ms glass-to-glass.  50 frames per second gives 1 frame as 20ms, meaning 5 fames is 100ms. Soliton latency over a wired connection is 35ms, within 2 video frames, which is well within any definition of ultra-low latency.

Does ultra-low latency for live streaming have other names?

It is sometimes referred as almost zero latency live streaming. This is implying that you are viewing the video in practically real-time. With a latency from Soliton over a wired connection being less than 35ms glass-to-glass, including camera , network latency and output monitor, this is less than 2 video frames and could be considered practically real-time.

What is the difference between low latency and ultra low latency?

Ultra Low Latency is video streaming from glass (camera) to glass (monitor) for example from  a remote location back to a studio, in less than 200ms. Low latency could be considered video streaming in less than 2 seconds glass-to-glass but there are no fixed definitions.

What are the drivers for ultra-low latency with live streaming?

Broadcasters are always seeking to lower latencies. For example by working with 5G standard bodies to investigate opportunities and developments such as network slicing and bandwidth frequency allocation to increase bandwidth, and thus seeking to lower latency and increase reliability for their video streaming workflows. However most broadcasters tend to be more focused on camera synchronization, i.e. fixed latencies on each camera for a multiple camera production, rather than ultra low latencies par se. A combination of both, within a managed cost framework, would be ideal. But ultra-low latency is not mission critical as distribution networks for the last mile, especially when playing out over the internet, can add considerable latency.

Is ultra low latency important?

Where ultra-low latency live streaming is really starting to get traction is in the newly emerging teleoperation and tele-driving market. This is defined as the remote control of untethered machines, robots, and vehicles in remote locations. If you are remotely driving a car via a live video stream, with a return remote control, while relying on public cellular for both control signals and video, even with a low latency of 0.5 seconds it would be extremely dangerous when pulling into traffic. Ultra-low latency within a couple of video frames is critical for safe remote control of machines and vehicles, even on unstable mobile phone networks.

Also remote surveillance in law enforcement and police applications recognize that any slight delay can have a massive impact on response and outcome, seconds really do count.

Mark Andrews

Subscribe Here!